
500 ◾ Software Metrics
to the successive failure time data of the Musa data, we generate the predic-
tions shown in the third column of Table 11.3.
In this case, the estimated distribution function is
Ft e
t
()=−
−
1
λ
where
λ
i
i
=
1
predictedmean timetothfailure
From this information, we can compute the sequence of u
i
s. For example,
uFtF e
1
16 5
30
30
84== =− =
−
() () .
.
Similarly, we compute the rest of the u
i
s as shown in the nal column of
Table 11.3.
Next, we construct the u-plot, shown in Figure 11.11, by placing the val-
ues of the u
i
s along the horizontal axis (which runs from 0 to 1) and then
drawing the step function where each step ...